<?php
// $Id: file_convert.module,v 1.27 2009/10/26 20:38:34 miglius Exp $

/**
 * @file
 * Allows files to be converted from one MIME content type into another as needed.
 */

//////////////////////////////////////////////////////////////////////////////

define('FILE_CONVERT_LIMIT_SIZE', (int)variable_get('file_convert_limit_size', 10));
define('FILE_CONVERT_DEBUG',      variable_get('file_convert_debug', 0));

define('FILE_CONVERT_TIER_LIMIT', 2);
define('FILE_CONVERT_DEBUG_LINES', 5);

//////////////////////////////////////////////////////////////////////////////
// Core API hooks

/**
 * Implementation of hook_help().
 */
function file_convert_help($path, $arg) {
  switch ($path) {
    case 'admin/settings/file/convert/converter':
      return '<p>'. t('Command line or php functions as utilities to convert between MIME types and create file previews of other MIME type. Empty field means that no conversion will take place. \'||\' separates utilities, which will be executed in sequence. Shell commands or php functios should be enclosed in {} brackets. The system will check if those functions are available or commands are executable. If there is a string and digits separated by the colon in the brackets like {localhost:8100} the system will check if the port on the host is opened. Other files which should be readable should be enclosed in double {{}} brackets. Brackets {}[] in command line can be escaped with \{\}\[\]. "No" in the "Installed" column means that not all commands and functions are found or network connection cannot be opened and the conversion will not take place.') .'</p>';
  }
}

/**
 * Implementation of hook_theme().
 */
function file_convert_theme() {
  return array(
    'file_convert_admin_converter_settings' => array(
      'arguments' => array('form' => NULL),
      'file' => 'file_convert.theme.inc'
    ),
  );
}

/**
 * Implementation of hook_perm().
 */
function file_convert_perm() {
  return array('re-run file conversions');
}

/**
 * Implementation of hook_menu().
 */
function file_convert_menu() {
  return array(
    'admin/settings/file/convert' => array(
      'title' => 'Converters',
      'page callback' => 'drupal_get_form',
      'page arguments' => array('file_convert_admin_settings'),
      'access arguments' => array('administer site configuration'),
      'file' => 'file_convert.admin.inc',
    ),
    'admin/settings/file/convert/configure' => array(
      'title' => 'Settings',
      'type' => MENU_DEFAULT_LOCAL_TASK,
    ),
    'admin/settings/file/convert/converter' => array(
      'title' => 'MIME Converters',
      'type' => MENU_LOCAL_TASK,
      'page callback' => 'drupal_get_form',
      'page arguments' => array('file_convert_admin_converter_settings'),
      'access arguments' => array('administer site configuration'),
      'weight' => 1,
      'file' => 'file_convert.admin.inc',
    ),
    'file/%node/conversion' => array(
      'title' => 'File conversion',
      'type' => MENU_CALLBACK,
      'page callback' => 'file_convert_rerun',
      'page arguments' => array(1, 3),
      'access callback' => 'node_access',
      'access arguments' => array('view', 1),
      'file' => 'file_convert.pages.inc',
    ),
  );
}

/**
 * Implementation of hook_link().
 */
function file_convert_link($type, $node = NULL, $teaser = FALSE) {
  $links = array();
  if (user_access('re-run file conversions') && $type == 'node' && $node->type == 'file' && !empty($node->file)) {
    $links['file_convert_conversion'] = array(
      'title' => t('Re-run conversions'),
      'href' => 'file/'. $node->nid .'/conversion/'. $node->vid,
      'attributes' => array('title' => t('Re-run file conversions for this file.')),
    );
  }
  return $links;
}

//////////////////////////////////////////////////////////////////////////////
// File API extensions

/**
 * This function executes all hook_mime_converters() hooks in the format modules
 * and creates an array of the MIME convert functions
 *
 * @return
 *   A structured array of the convert functions.
 */
function file_get_mime_converters() {
  static $converters = array();
  if (empty($converters)) {
    $saved_converters = array();
    foreach (module_implements('mime_converters') as $module) {
      if (($types = module_invoke($module, 'mime_converters')) && is_array($types)) {
        foreach ($types as $from => $tos) {
          foreach ($tos as $to => $converter) {
            if (!isset($converter['handlers']))
              $types[$from][$to]['handlers'] = array();
            $types[$from][$to]['module'] = $module;
            $types[$from][$to]['enabled'] = 1;
          }
        }
        $converters = array_merge_recursive($converters, $types);
      }
      $saved_converters = array_merge($saved_converters, variable_get($module .'_convert_pipelines', array()));
    }
    drupal_alter('mime_converters', $converters);
    foreach ($converters as $from => $tos) {
      foreach ($tos as $to => $converter) {
        if (isset($saved_converters[$from][$to])) {
          $converters[$from][$to]['pipeline'] = $saved_converters[$from][$to]['pipeline'];
          $converters[$from][$to]['enabled'] = $saved_converters[$from][$to]['enabled'];
        }
        $converters[$from][$to]['installed'] = _file_convert_check_pipeline($converters[$from][$to]['pipeline']);
        if (empty($converters[$from][$to]['installed'])) {
          $converters[$from][$to]['enabled'] = '';
        }
      }
    }
  }
  return $converters;
}

/**
 * Converts a file to different MIME format.
 * If earlier conversion failed, the file is not converted.
 *
 * @param $file
 *   A file object. If $file->converted defined, we already have a temporal file.
 * @param $handlers
 *   An array of the handlers to be used with the converted file.
 * @param $pipeline
 *   An array with a pipelines.
 * @param $options
 *   Additional options used in file conversion.
 *
 * @return
 *   A converted file's filename or NULL.
 */
function file_convert($file, $pipeline, $tier, $handlers = array(), $options = array()) {
  $status = TRUE;
  foreach ($handlers as $handler) {
    $status = $status && !_file_convert_isnot_needed($file, $handler);
  }
  if (!$status)
    return NULL;

  $status = _file_convert($file, $pipeline, $tier, $options);
  if (!isset($status) || $status === FALSE) {
    foreach ($handlers as $handler) {
      _file_convert_failure_log($file, $handler);
    }
  }

  return $status;
}

/**
 * Converts a file to different MIME format.
 *
 * @param $file
 *   A file object. If $file->converted defined, we already have a temporal file.
 * @param $pipeline
 *   An array with a pipelines.
 * @param $options
 *   Additional options used in file conversion.
 *
 * @return
 *   A converted file's filename or NULL. Calling function
 *   should delete the temporal file.
 */
function _file_convert($file, $pipeline, $tier, $options) {
  $options['filename'] = preg_replace('/"/', '\"', $file->filename);
  $options['tmpdir'] = file_directory_temp();
  $file_convert = ($tier > 0) ? $file->converted[$tier - 1] : $file;
  if (isset($file_convert->filepath)) {
    $in_file = $file_convert->filepath;
    $keep_in_file = TRUE;
  }
  else {
    // Copy file to the temporal file.
    $input = bitcache_get_stream(file_get_hash($file_convert->uri));
    $tmpfile = tempnam($options['tmpdir'], 'drupal_file_convert_');
    $output = fopen($tmpfile, 'wb');
    stream_copy_to_stream($input, $output);
    fclose($input);
    fclose($output);
    $in_file = $tmpfile;
    $keep_in_file = FALSE;
  }

  if (!_file_convert_check_pipeline($pipeline)) {
    watchdog('file_convert', 'Utilities for the pipeline %pipeline are not installed.', array('%pipeline' => $pipeline), WATCHDOG_ERROR, isset($file->nid) ? l(t('view'), 'node/'. $file->nid) : NULL);
    if (!$keep_in_file)
      file_delete($in_file);
    return FALSE;
  }
  foreach (explode('||', $pipeline) as $pipe) {
    if (isset($out_file)) {
      // Second and following pipes.
      if (!$keep_in_file)
        file_delete($in_file);
      $keep_in_file = FALSE;
      $in_file = $out_file;
    }
    $out_file = tempnam($options['tmpdir'], 'drupal_file_convert_');
    if (!_file_convert_execute_pipeline($file, $in_file, $out_file, $pipe, $options)) {
      watchdog('file_convert', 'File %name, uri=%uri was not converted by executing a pipeline %pipeline. Execution of the pipeline failed.', array('%name' => $file->filename, '%uri' => $file->uri, '%pipeline' => $pipe), WATCHDOG_ERROR, isset($file->nid) ? l(t('view'), 'node/'. $file->nid) : NULL);
      if (!$keep_in_file)
        file_delete($in_file);
      file_delete($out_file);
      return FALSE;
    }
  }
  if (!$keep_in_file)
    file_delete($in_file);

  if (filesize($out_file)) {
    watchdog('file_convert', 'File %name, uri=%uri was converted by executing a pipeline %pipeline.', array('%name' => $file->filename, '%uri' => $file->uri, '%pipeline' => $pipeline), WATCHDOG_NOTICE, isset($file->nid) ? l(t('view'), 'node/'. $file->nid) : NULL);
    return $out_file;
  }
  else {
    watchdog('file_convert', 'File %name, uri=%uri was not converted by executing a pipeline %pipeline. Converted file was empty.', array('%name' => $file->filename, '%uri' => $file->uri, '%pipeline' => $pipeline), WATCHDOG_ERROR, isset($file->nid) ? l(t('view'), 'node/'. $file->nid) : NULL);
    file_delete($out_file);
    return FALSE;
  }
}

//////////////////////////////////////////////////////////////////////////////
// Auxiliary functions

/**
 * Checks availability of the particullar command.
 */
function _file_convert_check_pipeline($pipeline) {
  foreach (explode('||', $pipeline) as $command) {
    if (!_file_convert_check_command($command)) {
      return NULL;
    }
  }
  // Check passed.
  return TRUE;
}

/**
 * Checks availability of the particullar command.
 */
function _file_convert_check_command($command) {
  if (trim($command) == '') {
    return FALSE;
  }

  // Escaping '{', '}', '[', ']'.
  $args = array('/\\\\{/', '/\\\\}/', '/\\\\\[/', '/\\\\\]/');
  $replacements = array('-left-curly-bracket-', '-right-curly-bracket-', '-left-square-bracket-', '-right-square-bracket-');
  $command = preg_replace($args, $replacements, $command);

  if (preg_match_all('/{([^}]+)}\(/', $command, $matches)) {
    // php function.
    foreach ($matches[1] as $function) {
      if (!function_exists($function)) {
        return FALSE;
      }
    }
  }
  else if (preg_match_all('/{([^}]+)}/', $command, $matches)) {
    // shell cyommand or library, or network connection.
    foreach ($matches[1] as $function) {
      if (preg_match('/{(.+)/', $function, $match)) {
        // a library.
        if (!is_readable($match[1])) {
          return FALSE;
        }
      }
      else if (preg_match('/(\S+):(\d+)/', $function, $match)) {
        // network connection.
        if (!($fp = @fsockopen($match[1], $match[2], $errno, $errstr, 10))) {
          return FALSE;
        }
        else {
          fclose($fp);
        }
      }
      else {
        // shell command.
        if (!_file_command_exists($function))
          return FALSE;
      }
    }
  }
  // Check passed.
  return TRUE;
}

/**
 * Executes a pipeline to convert a file.
 */
function _file_convert_execute_pipeline($file, $from, $to, $pipeline, $options) {
  $result = TRUE;

  $debug = array(
    'pipeline' => $pipeline,
    'from' => $from,
    'to' => $to,
  );

  // Escaping '{', '}', '[', ']'.
  $args = array('/\\\\{/', '/\\\\}/', '/\\\\\[/', '/\\\\\]/');
  $replacements = array('-left-curly-bracket-', '-right-curly-bracket-', '-left-square-bracket-', '-right-square-bracket-');
  $pipeline = preg_replace($args, $replacements, $pipeline);

  $args = array('/{\S+:\d+}/', '/{{/', '/}}/', '/{/', '/}/', '/\[in_file\]/', '/\[out_file\]/', '/\[in_file_base\]/', '/\[out_file_base\]/', '/\[in_content\]/', '/\[out_content\]/');
  $replacements = array('', '', '', '', '', $from, $to, basename($from), basename($to), '$in_content', '$out_content');
  foreach ($options as $name => $val) {
    array_push($args, '/\['. $name .'\]/');
    array_push($replacements, $val);
  }
  $pipeline = preg_replace($args, $replacements, $pipeline);

  // Putting back escaped values.
  $args = array('/-left-curly-bracket-/', '/-right-curly-bracket-/', '/-left-square-bracket-/', '/-right-square-bracket-/');
  $replacements = array('{', '}', '[', ']');
  $pipeline = preg_replace($args, $replacements, $pipeline);
  $debug['pilpeline_expanded'] = $pipeline;

  if (preg_match('/\$in_content/', $pipeline)) {
    // Read content from file.
    $in_content = file_get_contents($from);
  }
  if (preg_match('/^.*\s([\w\d_]+)\(.*$/', ' '. $pipeline, $matches)) {
    // php function.
    if (!eval($pipeline .'; return 1;')) {
      watchdog('file_convert', 'Execution of pipeline %pipeline failed.', array('%pipeline' => $pipeline), WATCHDOG_ERROR);
      $result = FALSE;
    }
    $debug['type'] = 'php';
  }
  else {
    // shell command.
    $exitcode = _file_command_run($pipeline, $pipes) ;
    if ($exitcode === FALSE) {
      watchdog('file_convert', 'Execution of pipeline %pipeline failed.', array('%pipeline' => $pipeline), WATCHDOG_ERROR);
      $result = FALSE;
    }
    $debug = array_merge($debug, array(
      'type' => 'shell',
      'stdout' => $pipes['stdout'],
      'stderr' => $pipes['stderr'],
      'exit_code' => $exitcode,
      'from_size' => filesize($from),
      'to_size' => filesize($to),
    ));
  }
  if (preg_match('/\$out_content/', $pipeline)) {
    // Save content to file.
    if (!file_put_contents($to, $out_content)) {
      watchdog('file_convert', 'Writting to a file %to failed.', array('%to' => $to), WATCHDOG_ERROR);
      $result = FALSE;
    }
  }
  $debug['result'] = $result && filesize($to) ? 'OK' : 'FAILED';
  if (FILE_CONVERT_DEBUG) {
    $out = array();
    foreach ($debug as $key => $val) {
      $out[] = '-- '. $key .' --';
      $val = trim($val);
      if (!empty($val))
        $out[] = implode("\n", array_slice(explode("\n", $val), -FILE_CONVERT_DEBUG_LINES));
    }
    watchdog('file_convert', 'Debug of the conversion:<br/>!out', array('!out' => implode('<br/>', $out)), $result ? WATCHDOG_NOTICE : WATCHDOG_ERROR);
  }
  $file->debug[] = $debug;
  return $result;
}

/**
 * Saves to the RDF that a conversion perfomed by the handler failed.
 * This will avoid continuous conversion of the failing file.
 *
 * @param $file
 *   A file object.
 * @param $handler
 *   A handler name.
 *
 * @return
 *   TRUE if save was successfull.
 */
function _file_convert_failure_log($file, $handler) {
  if (!rdf_exists($file->uri, 'wordnet:failed', $handler, array('repository' => FILE_RDF_REPOSITORY)))
    return rdf_insert($file->uri, 'wordnet:failed', $handler, array('repository' => FILE_RDF_REPOSITORY));

  return TRUE;
}

/**
 * Checks the RDF if the conversion was already made or had failed.
 *
 * @param $file
 *   A file object.
 * @param $handler
 *   A handler name.
 *
 * @return
 *   TRUE if failure is logged.
 */
function _file_convert_isnot_needed($file, $handler) {
  if (rdf_exists($file->uri, 'wordnet:failed', $handler, array('repository' => FILE_RDF_REPOSITORY)))
    return TRUE;

  if ($generated = rdf_normalize(rdf_query(NULL, rdf_qname_to_uri('dc:source'), $file->uri, array('repository' => FILE_RDF_REPOSITORY)))) {
    foreach ($generated as $uri => $data) {
      $preview = rdf_normalize(rdf_query($uri, rdf_qname_to_uri('dc:creator'), $handler, array('repository' => FILE_RDF_REPOSITORY)));
      if (!empty($preview))
        return TRUE;
    }
  }

  return FALSE;
}